Power over ethernet

Today is my day off so I had a little time to piddle. After seeing a
couple of mods on the internet to put POE on a WRT54 I decided to
build my own. The mods I saw involved hacking the box and I was not
fond of this idea since especially since my case started to crack when
I tried to get the halves apart. I picked up a couple of double wall
connections from the Home Depot . I connected the #1 jack to the #2
jack on the same outlet except for the wires used for power. On these
I connected the output of my wall wart power suppy.

On the other end I did the same connecting the power plug removed
from same wall wart to the power leads on #1 jack. Realizing someone
may cross up some wires and blow up their router like this I tried to
do a little idiot proofing. AT the box used at the router I connected
a little bridge rectifier removed from a modem board with the AC
terminals gong to the line and the + and _ terminals going to the
appropriate terminals on the power jack.

Yes I am being intentionally vague with these directions. Those with
the appropriate amount of electronic savy should have no problems.
This should work for about any wall wart powered device.

jimmie68@gmail.com wrote:
> Today is my day off so I had a little time to piddle. After seeing a
> couple of mods on the internet to put POE on a WRT54 I decided to
> build my own.

Way too complicated. What I've done several times when I couldn't get
AC power to the Linksys, I ran a "poor man's PoE" with splicing the
power leads in to and out of the CAT5 cable at both ends.

Takes only five minutes and a few beenies (white telephones connectors),
but regular crimp connectors would work.

DTC wrote:
> jimmie68@gmail.com wrote:
>> Today is my day off so I had a little time to piddle. After seeing a
>> couple of mods on the internet to put POE on a WRT54 I decided to
>> build my own.
>
> Way too complicated. What I've done several times when I couldn't get
> AC power to the Linksys, I ran a "poor man's PoE" with splicing the
> power leads in to and out of the CAT5 cable at both ends.

Yup, I ran an Actiontec router like that in my conservatory for some years.

> Takes only five minutes and a few beenies (white telephones connectors),
> but regular crimp connectors would work.

I did it with some insulating tape.... its only 5v.

On Tue, 15 Jul 2008 22:38:58 +0100, Mark McIntyre
<markmcintyre@TROUSERSspamcop.net> wrote:

>I did it with some insulating tape.... its only 5v.

Only the WRT54G v1 and *SOME* v1.1 versions used 5VDC power. I never
could get one to work through a mess of CAT5 because the router
required exactly 5.0000VDC to work correctly and the relatively high
current through the copper produced excessive voltage drops. Later
WRT54G versions were supplied with 12V 1A wall warts, and included a
wide range regulator that works from about 4.0VDC to 18VDC. Those
worked very well with the method you detailed. I've done much the
same thing, but without the electrical tape. I use shrink tube.

Ever wonder why real PoE uses 48VDC? It's to reduce the current
through the CAT5 cable and RJ45 connectors. At a constant power
level, running something at 5VDC requires approximately 10 times the
current of running it at 48VDC. Voltage drop and heating are related
to current, not applied voltage. Also, the RJ45 connectors are rated
at 1A max per pin. The WRT54G will not exceed the 1A max rating, but
higher current devices might do so.

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Mark McIntyre wrote:
> DTC wrote:
> > jimmie68@gmail.com wrote:
> >> Today is my day off so I had a little time to piddle. After seeing a
> >> couple of mods on the internet to put POE on a WRT54 I decided to
> >> build my own.
> >
> > Way too complicated. What I've done several times when I couldn't get
> > AC power to the Linksys, I ran a "poor man's PoE" with splicing the
> > power leads in to and out of the CAT5 cable at both ends.
>
> Yup, I ran an Actiontec router like that in my conservatory for some years.
>
> > Takes only five minutes and a few beenies (white telephones connectors),
> > but regular crimp connectors would work.
>
> I did it with some insulating tape.... its only 5v.

Yep you can do that but you can also damage a device that way if you
are not careful. I was trying to come up with something a little more
idiot proof. An example, recently I showed my friend how I added an
external battery pack to my camera. A few days later he is complaining
to me that he destroyed his camera like it is my fault he wired the
thing up backwards.

Jimmie

Jimmie

Jeff Liebermann wrote:
> On Tue, 15 Jul 2008 22:38:58 +0100, Mark McIntyre
> <markmcintyre@TROUSERSspamcop.net> wrote:
>
> >I did it with some insulating tape.... its only 5v.
>
> Only the WRT54G v1 and *SOME* v1.1 versions used 5VDC power. I never
> could get one to work through a mess of CAT5 because the router
> required exactly 5.0000VDC to work correctly and the relatively high
> current through the copper produced excessive voltage drops. Later
> WRT54G versions were supplied with 12V 1A wall warts, and included a
> wide range regulator that works from about 4.0VDC to 18VDC. Those
> worked very well with the method you detailed. I've done much the
> same thing, but without the electrical tape. I use shrink tube.
>
> Ever wonder why real PoE uses 48VDC? It's to reduce the current
> through the CAT5 cable and RJ45 connectors. At a constant power
> level, running something at 5VDC requires approximately 10 times the
> current of running it at 48VDC. Voltage drop and heating are related
> to current, not applied voltage. Also, the RJ45 connectors are rated
> at 1A max per pin. The WRT54G will not exceed the 1A max rating, but
> higher current devices might do so.
>


PoE seems to be used a lot in HVAC and Alarm systems now that a lot of
the controls and sensors are LAN based. I havent seen PoE used
extensiely in any other way. Jeff, as always your accurate and
thoughtful comments are always appreciated.

I had offered my way of hooking it up because I was thinking of
mounting a WRT54 in the cupola on my garage. I had seen methods on the
internet that suggested the router be opened up and diodes be placed
inside the router and connections to the power supply be made inside
the router. I had also coome across the simple methods suggested by
others here in this thread. The first I considered way too much
trouble and the second way too dependent on me wiring it up and not
goofing something up. Its posible the router may already have some
reverse polarity protection, even if this is the case I dont want to
have to troubleshoot the reason the router isnt working. By using a
small bridge rectifier at the router end it allows me to connect the
power supply at the other end with either polarity reducing the reason
my setup could fail. The extra few minutes I spent building my adapter
is insurance against an extra trip into my attic.


Jimmie
Jimmie

jimmie68@gmail.com wrote:
>
>>
> Yep you can do that but you can also damage a device that way if you
> are not careful.

I doubt it - worst case you'll have too low a voltage at the head end.
You /could/ damage your CAT5 mind you, if you had too much current.

> I was trying to come up with something a little more
> idiot proof. An example, recently I showed my friend how I added an
> external battery pack to my camera.

I make a point of never showing actual friends how clever I am with
stuff like that, unless I already know they're technically savvy.

>A few days later he is complaining
> to me that he destroyed his camera like it is my fault he wired the
> thing up backwards.

Hmm. Consumer electronics goods are pretty immune to people putting the
batteries in backwards.

Jeff Liebermann wrote:
> On Tue, 15 Jul 2008 22:38:58 +0100, Mark McIntyre
> <markmcintyre@TROUSERSspamcop.net> wrote:
>
>> I did it with some insulating tape.... its only 5v.
>
> Ever wonder why real PoE uses 48VDC?

Nope! I did a physics degree...

> to current, not applied voltage. Also, the RJ45 connectors are rated
> at 1A max per pin.

My technique is to cut the wall-wart cable in two, pull the unused two
pairs out of the cat5, and splice the two power lines into the pairs.
This avoids any current flow through the RJ45s. I've no idea what the
rating of cat5 is but a pair (still twisted) seemed ok to carry an amp.
Cable didn't get warm anyway....

On Thu, 17 Jul 2008 09:59:35 +0100, Mark McIntyre
<markmcintyre@TROUSERSspamcop.net> wrote:

>Jeff Liebermann wrote:
>> On Tue, 15 Jul 2008 22:38:58 +0100, Mark McIntyre
>> <markmcintyre@TROUSERSspamcop.net> wrote:
>>
>>> I did it with some insulating tape.... its only 5v.
>>
>> Ever wonder why real PoE uses 48VDC?
>
>Nope! I did a physics degree...

I thought physicists ended all their correspondence with "More
research is necessary".

>> to current, not applied voltage. Also, the RJ45 connectors are rated
>> at 1A max per pin.
>
>My technique is to cut the wall-wart cable in two, pull the unused two
>pairs out of the cat5, and splice the two power lines into the pairs.
>This avoids any current flow through the RJ45s. I've no idea what the
>rating of cat5 is but a pair (still twisted) seemed ok to carry an amp.
>Cable didn't get warm anyway....

That will work, but the less disgusting looking method is to use an
RJ45 wall "muffin" (connector block) for interconnect.

It's not cable heating that limits operation over CAT5 or what
inspired the 48VDC standard for PoE. It's the voltage drop that the
CAT5 cable presents. A single #24 awg wire in the CAT5 bundle has a
resistance of about 9.4 ohms per 100 meters (the official maximum for
802.11af PoE). If your WRT54G v1.0 required about 1.5A of current,
and you used 4 wires to carry the current, 100 meters of CAT5, the
delivered voltage to the WRT54G would be:
Loop resistance = 9.4 ohms
E = I * R = 1.5A * 9.4 ohms = 14.1 V
which is more than the applied 5VDC. That means more power is
dissipated in the wire than in the WRT54G. That won't work at all.

If I shorten the cable to 20 meters, I get:
Loop resistance = 20/100 * 9.4 = 1.9 ohms
E = I * R = 1.5 * 1.9 = 2.9 volts
Voltage delivered = 5.0VDC - 2.9VDC = 2.1VDC
I don't think the WRT54G v1.0 will run on 2.1VDC. However, if you
replace the 5VDC wall wart, with a 7.1VDC wall wart to compensate for
the drop, you will get about 5VDC at the radio, and it will work.

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

> Hmm. Consumer electronics goods are pretty immune to people putting the
> batteries in backwards.

For ones that use batteries, perhaps. But for something intended only for
use with a power adapter it's still potentially a risk.

Jeff Liebermann wrote:
>
> I thought physicists ended all their correspondence with "More
> research is necessary".

I did - in layman's terms "the cable didn't get warm, anyway..." surely
indicates further research necessary? :-) See my sig for further info.....

> It's not cable heating that limits operation over CAT5 or what
> inspired the 48VDC standard for PoE. It's the voltage drop that the
> CAT5 cable presents.

Well, both can be factors. But you're right - if you want to get 12V out
of the far end of a 100mtr cable, you'll need to stuff in way more than
12V at the near end.

> Voltage delivered = 5.0VDC - 2.9VDC = 2.1VDC
> I don't think the WRT54G v1.0 will run on 2.1VDC.

Luckily I wasn't using a WRT54G... :-)

I was using an actiontec wireless home gateway. My run was about (quick
sum) 100ft. I'm assuming that the vendor-provided wallwart chucked out
more voltage than the router really needed.

--

"Give me a couple of years and a large research grant,
and I'll give you a receipt." --Richard Heathfield

Bill Kearney wrote:
>> Hmm. Consumer electronics goods are pretty immune to people putting
>> the batteries in backwards.
>
> For ones that use batteries, perhaps. But for something intended only
> for use with a power adapter it's still potentially a risk.

Quite possibly - I read it that the "external battery pack" was being
attached to a battery-powered camera. I can't imagine a mains-powered
camera would be much use....

--
Mark McIntyre

CLC FAQ <http://c-faq.com/>
CLC readme: <http://www.ungerhu.com/jxh/clc.welcome.txt>

On Thu, 17 Jul 2008 20:23:51 +0100, Mark McIntyre
<markmcintyre@TROUSERSspamcop.net> wrote:

>Jeff Liebermann wrote:
> >
>> I thought physicists ended all their correspondence with "More
>> research is necessary".
>
>I did - in layman's terms "the cable didn't get warm, anyway..." surely
>indicates further research necessary? :-) See my sig for further info.....

Well, yes. It's a stretch but certainly within acceptable limits for
soliciting new projects.

>> It's not cable heating that limits operation over CAT5 or what
>> inspired the 48VDC standard for PoE. It's the voltage drop that the
>> CAT5 cable presents.
>
>Well, both can be factors. But you're right - if you want to get 12V out
>of the far end of a 100mtr cable, you'll need to stuff in way more than
>12V at the near end.
>
>> Voltage delivered = 5.0VDC - 2.9VDC = 2.1VDC
>> I don't think the WRT54G v1.0 will run on 2.1VDC.
>
>Luckily I wasn't using a WRT54G... :-)

Now you tell me.

>I was using an actiontec wireless home gateway. My run was about (quick
>sum) 100ft. I'm assuming that the vendor-provided wallwart chucked out
>more voltage than the router really needed.

Good assumption, but with two possible problems. Some devices (i.e.
Buffalo wireless routers) use 3.3VDC and 5.0VDC wall warts with
internal regulators. There's no regulator inside the box and the
chips run directly from the wall wart power. If you apply to high a
voltage, it smokes. Too low, and it won't play. The easy test is to
measure the wall wart voltage, with no load, using a DVM. If it's
exactly 3.3 or 5.0VDC, forget about trying your method. If the
voltage is somewhat higher (up to perhaps twice as high), the
regulator is inside the box and you can probably run it through the
CAT5 extension cord. Extra credit if you test it with a variac
(adjustable 117VAC xformer).


--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Jeff Liebermann wrote:
> On Thu, 17 Jul 2008 20:23:51 +0100, Mark McIntyre
> wrote
>> Luckily I wasn't using a WRT54G... :-)
>
> Now you tell me.

Actually, I said it much earlier in the thread. I suspect youre muddling
me with the OP.

>> I was using an actiontec wireless home gateway. My run was about (quick
>> sum) 100ft. I'm assuming that the vendor-provided wallwart chucked out
>> more voltage than the router really needed.
>
> Good assumption, but with two possible problems.

Absolutely. However I did measure the voltage at the head end to see if
it was sensible. I forget what it was, this /was/ a couple of years back.

> measure the wall wart voltage, with no load, using a DVM. If it's
> exactly 3.3 or 5.0VDC, forget about trying your method.

If I found a wallwart sold with a <£50 router that generated exactly any
voltage, I'd send it to the Smithsonian. It'd be a miracle of modern
engineering.... :-)

> Extra credit if you test it with a variac
> (adjustable 117VAC xformer).

Round here, the voltage is a pretty steady 240v.

--
Mark McIntyre

CLC FAQ <http://c-faq.com/>
CLC readme: <http://www.ungerhu.com/jxh/clc.welcome.txt>